Everything in AgentsOS.

A complete AI operating system spanning orchestration, execution, memory, routing, and business verticals. Every component ships together as one platform.

User
Hermes Orchestrator (8039)
↓ Task Classifier
Gemini Flash / GPT-5 nano
↓ Routes to
Coding → GPT-5.5 / Claude Opus Reasoning → o3 / GPT-5.5 Vision → Qwen-VL / Gemini Vision Local → Ollama (Qwen3, Gemma, DeepSeek) Fast → Gemini Flash
↓ Provider Layer
LiteLLM Proxy 9Router (fallback)
↓ Providers
OpenAI Anthropic OpenRouter Ollama Azure vLLM

AgentsOS Core

SHIPPED v3.1 Runtime FastAPI Python

The runtime platform: agent execution engine, DAG orchestration, event bus, model routing (local/cloud), and the core API surface. Everything in AgentsOS runs on top of this.

Status
✅ Shipped v3.1 — 80+ tests, production-ready
Components
Control Plane (8080), DAG Engine, Event Bus, PDO Models
Stack
Python FastAPI, Pydantic, SQLAlchemy, pytest, ruff
Next
PostgreSQL migration, JWT auth, durable task queue (v3.2)

Master Orchestrator

BETA v3.1 Planning DAG Scheduling

Takes a project brief, plans it as a DAG, resources it against the agent fleet, executes it end-to-end, reviews results, and learns from every outcome. The core intelligence of the platform.

Lifecycle
Brief → PDO → Task Plan → DAG → Execute → Review → Learn
Hermes Engine
DAG planning & scheduling (port 8039)
Safety Modes
YOLO, Review, Approval — per-task gating
Coming in v4.0
LLM-native hierarchical planner, evolution wiring, provenance-gated execution

Agent Fleet

SHIPPED 6 modalities Concurrent

Six specialist agents, each with its own modality, load tracking, and concurrency model. The fleet scales per-core with auto-tuned headroom.

🖥️ Code Agent
Code generation, editing, refactoring
☁️ Infra Agent
Docker, Cloudflare, Tailscale, cloud console
👁️ Vision Agent
Screenshot analysis, diagram understanding, OCR
🌐 Browser Agent
Playwright-based web automation (Phase 2)
📱 Android Agent
Device control via ADB + Hermes Bridge (Phase 2–3)
📄 Docs Agent
RAG-powered document read/write/summary

Desky

BETA Chat Project Director

Your conversational project director. Desky lives in the MyDesk dashboard, speaks DAG, generates briefs, and guides you through project lifecycle from kickoff to lessons learnt.

Capabilities
DAG preview, approval gates, multi-turn planning
Integration
MyDesk frontend persona — cosmetic layer, no infra
Modes
Chat, Brief Generator, Project Review, Lessons Learnt
Coming
Desky Pro — multi-project portfolio view (v4.1)

OpenClaw

SHIPPED v1.0 Executor Memory Provenance

The step executor and memory graph. OpenClaw runs individual task nodes using a fixed set of primitives (run_command, read/write, http_request, model_infer), writes every action to the provenance ledger, and validates results against acceptance criteria.

Port
4096
Ledger
Every action writes to Neo4j provenance graph
Primitives
run_command, read/write, http_request, model_infer, run_python, run_node
Coming in v2
Sandboxed execution, rollbacks, retry policies (v4.0)

Model Router

BETA Routing Fallback Multi-Provider

Three-layer routing architecture: Agent Router (which specialist) → Model Router (which LLM) → Provider Router (which endpoint). Task-classifier-driven with automatic fallback chains.

Layer 1 — Agent
Hermes classifies intent → routes to Code/Infra/Vision/Browser/Android/Docs agent
Layer 2 — Model
Task type determines model: coding→GPT-5.5, reasoning→o3, vision→Qwen-VL, fast→Gemini Flash
Layer 3 — Provider
LiteLLM → 9Router fallback → cheapest/healthiest endpoint
9Router
122 models, 40+ providers, automatic tiered fallback (subscription → cheap API → free). Available as Docker profile.

MCP Tool Marketplace

SHIPPED Registry Plugins

A registry of execution tools and skills that agents can discover and invoke. MCP-compatible connectors expand the platform's reach into any external system.

Port
7070
Connectors
GitHub, Cloudflare, Tailscale, Docker, Slack, Teams
Discovery
Agents query registry for available tools by capability tag
Extensible
Any MCP-compatible server can register itself

Enterprise Brain

BETA Memory RAG Graph

The knowledge and memory layer. Qdrant provides semantic/vector retrieval (RAG) over docs and code. Neo4j stores the agent activity ledger and capability/relationship graph. Redis handles short-term session memory and task queues.

Qdrant (6333)
Vector store — RAG over code, docs, customer data
Neo4j (7474/7687)
Graph store — provenance ledger, agent relationships
Redis (6379)
Queue, cache, session memory
Coming
Digital Twin — personalisation layer over ledger (Phase 3)

MyDesk AI

BETA Dashboard Workspace Blazor

The human-facing workspace: agent dashboard, chat, DAG visualisation, and business verticals (legal, finance, CRM). Consumes AgentsOS over the Control Plane API exclusively — never talks to agents directly.

Stack
Blazor / MudBlazor 7.15, .NET 10, Next.js dashboard shell
Verticals
Legal (AU Consumer Law, QLD Legal Practice Act) — pilot ready
Controls
Project dashboard, DAG visualisation (D3.js), approval gates
Integration
REST + WebSocket to AgentsOS API. Never owns orchestration logic.

Hermes Orchestrator

SHIPPED Agent Model Switching Provider Abstraction

First-party AI orchestration agent. Hermes handles model switching via hermes model, provider abstraction (OpenAI/Anthropic/Ollama/vLLM/OpenRouter), gateway routing, and serves as the DAG planning engine inside Master Orchestrator.

Port
8039
Auto-Routing
Intent classifier → role-based session routing (RFC)
Fallback
Automatic provider failover, credential pools
Integration
Bridged with AgentsOS Docker stack. Windows pb-legacy ↔ Docker.
ℹ️ Hermes Routing Architecture — The platform uses a three-layer routing model: Agent Router (which specialist agent) → Model Router (which LLM) → Provider Router (which endpoint). This avoids combining all decisions into one monolithic router and follows the industry pattern emerging across 9Router, LiteLLM, and the Hermes Auto-Routing RFC.

9Router Gateway

SHIPPED Fallback Subscriptions

Optional AI gateway for provider-level failover and cost optimisation. 9Router sits behind LiteLLM as a fallback layer, managing subscription tiers, API quotas, and automatic failover across 40+ providers.

Status
✅ Running on localhost:20128 — 122 models
Docker Profile
docker compose --profile 9router up -d
Skills
8 Hermes skills installed (chat, image, tts, stt, embeddings, web-search, web-fetch)
Tiered Fallback
Subscription → Cheap API → Free providers, no user interaction

What's coming next.

The platform evolves fast. This roadmap is updated with every release — check back often.

✅ DONE

Phase 1 — Foundation Hardening

v3.1
  • Control Plane FastAPI — 80+ tests passing
  • DAG Engine with async task lifecycle
  • PDO Models (Scope, Risk, Epic, TaskNode, ResourcePlan)
  • Master Orchestrator — planning → execution loop
  • Evolution Engine — DAGGenome, fitness scoring, genetic optimization
  • Event Bus — async pub/sub
  • Frontend (Next.js + Tailwind + D3.js)
  • Docker Compose full stack (15 services)
  • 9Router AI Gateway integration with 8 Hermes skills
  • 5 codebase duplicates consolidated → /archive
  • Marketing site rebuilt matching agents-os-orchestra.base44.app
  • Three-layer routing architecture (Agent → Model → Provider)
🔄 IN PROGRESS

Phase 2 — Persistence & Auth

v3.2
  • Database migration — in-memory state → PostgreSQL
  • AuthN/AuthZ — JWT, tenant isolation, RBAC
  • Secrets rotation — remove hardcoded secrets from docker-compose
  • Durable task queue — Redis-backed
  • Activity ledger — write-audit trail to Neo4j
  • API versioning — /v1/ prefix on all routes
  • Android agent ADB service — persistent device connection
  • Hermes ↔ AgentsOS bridge — Windows pb-legacy ↔ Docker
🔮 PLANNED

Phase 3 — Orchestrator v2

v4.0 — Q4 2026
  • Hermes v2 — LLM-native hierarchical planner (replaces static decompose_task)
  • OpenClaw v2 — sandboxed execution, rollbacks, retry policies
  • Evolution wiring — GeneticEngine connected to PDO corpus
  • State machine — formal node lifecycle (pending→running→completed→failed→rolled_back)
  • Provenance-gated execution — no credit without ledger entry
  • Android agent full-cycle — brief → screen → plan → execute → verify
  • Multi-device Android management
🔮 PLANNED

Phase 4 — Monitoring & Observability

v4.1 — Q4 2026
  • Grafana dashboards — tokens, latency, cost, GPU, agents
  • Prometheus /metrics on all services
  • Structured JSON logging with correlation IDs
  • Slack/PagerDuty alerting for failures
  • Android device telemetry
🔮 PLANNED

Phase 5 — Enterprise Features

v4.2 — Q1 2027
  • SOC 2 compliance — access controls, audit logs, data retention
  • Multi-tenant isolation — per-org data, queues
  • SSO / SAML — Okta, Azure AD, Google Workspace
  • On-prem appliance — air-gapped deployment
  • Android compliance — device policy, screen recording audit, remote wipe
🔮 PLANNED

Phase 6 — Business Verticals

v4.3+ — Rolling
  • Legal (Australia) — AU Consumer Law, QLD Legal Practice Act — pilot ready
  • Financial Services — SOX compliance, reconciliation, AML checks
  • Healthcare — HIPAA, PHI handling, medical coding
  • Government — IRAP, PROTECTED, air-gapped deployment

Get early access.

Join the private beta. We're onboarding enterprise and pilot customers now.

No commitment. We'll review your application within 48 hours.